Inside Gong’s AI Strategy: Context, Memory, and Human Trust

Noaa Ilani, VP of Product at Gong, is helping redefine how AI fits into the sales stack — not as a replacement for reps, but as the intelligence and automation layer that sits on top of a decade of rich go-to-market data. She shares how Gong built a deep context layer long before LLMs, why trust and relationships still sit at the center of every deal, and how AI can quietly eliminate the 70% of work sellers do that isn’t actually selling.

Where AI Selling Ends — and Human Trust Begins

• AI SDRs for the Long Tail: Automated outreach can uncover opportunities in colder, long-tail accounts that humans never have time to touch.

• Humans Buy From Humans: For core, high-value deals, buyers still want to build trust with a real person — AI can be in the loop, but not the whole loop.

• Augment, Don’t Replace: The sweet spot is using AI to catch what humans miss, not to stand in for the relationship itself.

Building a Context Layer for AI-First Selling

• Data Before Models: Gong’s edge starts with a decade of calls, emails, and deal activity organized into a structured context layer.

• Beyond One Call: Deep research lets reps query months of conversations and activity across many deals, not just a single interaction.

• Themes You’d Never See Alone: Gong’s AI surfaces patterns across accounts and segments that would be impossible to spot manually.

Designing AI That Understands Real Conversations

• LLMs + Proprietary Models: Early on, Gong combined its own models with LLMs and “smart trackers” to highlight key phrases and moments.

• Guiding the Model’s Attention: System prompts and trackers steer AI toward what matters — not small talk or irrelevant side paths.

• Evolving With Better Models: As frontier models improve, Gong is simplifying prompts and giving customers more control over how AI is configured.

Meeting Sellers Wherever They Work

• Platform and Embedded: Some teams now spend their entire day in Gong; others stay in their CRM and bring Gong data to where they already live.

• Workflows Across the Stack: Extensions in tools like Gmail and CRM let sellers access Gong’s intelligence without changing their habits overnight.

• Plugging Into Agent Ecosystems: Gong is launching an MCP so its data and insights can power generalist agents like ChatGPT and Claude securely.

From Signals to Agentic Workflows

• More Than Call Data: Gong combines conversations with external sources like NPS, support tickets, and news to generate richer signals.

• Weighted Alerts That Matter: Multiple weak signals — a low NPS, support friction, and call mentions — roll up into clear, actionable triggers.

• Agents That Don’t Miss a Beat: Agentic workflows can continuously scan for risk and opportunity, surfacing what busy humans would otherwise overlook.

Reimagining How Sellers Spend Their Time

• 70% of Time Not Selling: Most seller hours still go to admin, updates, and follow-ups — work that’s critical but rarely fulfilling.

• From Admin to Outcomes: Gong’s vision is to shrink that non-selling time dramatically through automation and AI-driven actions.

• Freeing Humans for Human Work: The goal isn’t full autonomy at all costs — it’s giving reps back the time and focus to build trust, solve problems, and win.

Why It Matters
The future of sales isn’t a fully automated AI rep — it’s a deeply informed human seller backed by an AI system that knows every interaction, every signal, and every shift in the account. Gong’s journey shows how to pair a decade of revenue data with modern AI so teams can sell with more context, more confidence, and a lot less busywork.